IIIT Students Develop System To Detect Hate Speech On Twitter

IIIT Students Develop System To Detect Hate Speech On Twitter
IIIT Students Develop System To Detect Hate Speech On Twitter

In what could be seen as a major development in cracking down on offensive content on social media, a group of students from Indian Institute of Information Technology(IIIT) has developed a system to detect hate speech on Twitter.

According to a report by The Times of India, the students were working on the topic of ' deep learning for hate speech' on Twitter since a year. The team  includes professor and dean(R & D) Vasudev Varma at IIIT-H alongwith students Pinkesh Badjatiya, Shashank Gupta and adjunct faculty Manish Gupta.

Advertisement opens in new window

The team members are claiming that the system can detect abusive language, racist and sexist speech and flag offensive content.

Quoting Vasudev Varma, the TOI report says that by using a popular approach called supervised learning a computer algorithm is fed many examples of text from each form of hate, which can be categorised as `racist' or `sexist' tweets.

The algorithm smartly analyses the data and recognises racism or sexism, informed Varma. He added that algorithm uses neural networks, more popularly called deep learning.

Inspired from the human brain, the algorithms try to simulate how humans learn from examples.

Earlier in 2015, Ethan Czahor, a chief technology officer working  for American businessman Jeb Bush, had created a app  that removes any posts that might cause  problems with current or future employer.

Users have to search for keywords such as ‘gay’, ‘black’ or swear words and the program deletes any tweet that mentions them, reported Mail Online.

Post a Comment

You are not logged in, To comment please / Register
or use
Next Story : Cow Vigilantes: Maneka Gandhi’s PFA Trained Suspects in Delhi Attack, Claims Organisation’s Haryana Chief
Download the Outlook ​Magazines App. Six magazines, wherever you go! Play Store and App Store


A B C D E F G H I J K L M N O P Q R S T U V W X Y Z 0 1 2 3 4 5 6 7 8 9

or just type initial letters